Citations

Abstract Background Accurate diagnosis and risk stratification in myeloproliferative neoplasms (MPNs) rely on expert interpretation of complex clinical records. Outside major academic centers, specialist expertise is scarce, and time-consuming chart reviews extend clinic visits and delay treatment decisions, underscoring the need for automated, point-of-care phenotyping for MPN patients. Methods Patients with the ICD codes of PV, ET and MF were identified from the electronic health record. Cases were defined as patients meeting the 2017 WHO diagnostic criteria for PV, ET or MF or those with a definitive MPN diagnosis determined by a hematologist. Controls were the patients who met neither the diagnostic criteria nor had a hematologist-assigned diagnosis. Two clinicians extracted demographic, clinical (symptoms, thrombotic history), laboratory (complete blood count, serum chemistry), genetic (mutations, karyotype), bone marrow morphology, diagnostic mentions and prognostic score information from unstructured clinical notes and pathology reports. Prognostic scores included the conventional risk model for PV, the International Prognostic Score of Thrombosis (IPSET) for ET and the Dynamic International Prognostic Scoring System (DIPSS), DIPSS-plus or Mutation-enhanced International Prognostic Scoring System (MIPSS-70) for MF. Discrepancies in data extraction were resolved by a consensus. A random sample of 510 patients (PV: 85, ET: 85, MF: 85, controls: 255) was split into prompt development (n=60) and test (n=450) sets for iterative workflow engineering and evaluation. The workflow included: (i) a rule-based algorithm for automated retrieval of demographic details, the first bone marrow biopsy report, the first clinical note after biopsy and the latest laboratory results within 90 days preceding biopsy (ii) a HIPAA-compliant LLM (GPT-4 turbo) for data extraction from the clinical notes and pathology reports (iii) a regular expressions-based algorithm for post-processing extracted data into structured tables, (iv) a clinician-informed source prioritization algorithm to select data from preferred sources, (v) a modified WHO-informed diagnostic characterization tool incorporating hematologist-assigned diagnoses to distinguish cases from controls and; (vi) an NCCN guidelines-based algorithm for prognostic tool selection and risk-stratification based on available data. Structured prompts were used to extract demographic, clinical, laboratory, genetic and pathological data in a zero-shot deterministic setting. The performance of automated data extraction, diagnostic characterization and prognostication was compared against the manually curated data. Accuracy and F1 scores were computed for data extraction and prognostication tasks while sensitivity and specificity were calculated to assess diagnostic performance. Results A total of 450 (PV: 150, ET: 150, MF: 150) pathology reports and 50 (PV: 52, ET: 55; MF: 65) clinical notes were identified for 450 test set patients. The test set data extraction accuracy was 95% (reports: 97%, notes: 92%) for PV, 93% (reports: 92%, notes: 94%) for ET and 98% (reports: 99%, notes 97%) for MF. The test set F1 score was 0.95 (reports: 0.96, notes: 0.93) for PV, 0.95 (reports: 0.95, notes: 0.96) for ET and 0.98 (reports: 0.99, notes: 0.97) for MF. PV patients were diagnosed with 100% sensitivity and 98% specificity whereas ET and MF patients were diagnosed with 100% sensitivity and specificity. The accuracy of risk stratification was 100% for PV and ET and 94% for MF. The F1 score for risk stratification was 1.00 for PV and ET and 0.94 for MF. Conclusions The first-in-class automated framework successfully identified pertinent clinical information, performed accurate data extraction from unstructured records and enabled clinician-guided source prioritization for accurate diagnosis and risk stratification of MPN patients. Prospective validation of the framework is required for widespread clinical implementation.

Abstract Background: As mounting numbers of patients receive allogeneic hematopoetic stem cell transplantation (allo-SCT) for hematologic malignancies, focus on survivorship and post-transplant complications is increasing. 30-60% of patients develop pulmonary complications, making the lungs the most affected organ post-transplant. While obstructive lung disease post-allo-SCT has been studied as a manifestation of graft-versus-host-disease (GVHD) in the form of bronchiolitis obliterans syndrome, many patients' post-transplant pulmonary function tests (PFTs) reveal restrictive patterns of disease which have been understudied. The pathophysiology, risk factors, and clinical course of restrictive lung disease (RLD) post-allo-SCT have not been fully elucidated. This retrospective study sought to characterize baseline characteristics, imaging findings, and possible risk factors of patients with RLD post-allo-SCT. Methods: The study included all patients receiving allo-SCT at our institution between 2005 and 2021 with at least one screening PFT with a restrictive pattern. Restriction is defined as normal forced expiratory volume in one second (FEV1) to forced vital capacity (FVC) ratio and total lung capacity (TLC) less than the lower limit of normal. Data collected via chart review included demographic information; indication for and type of transplant; myeloablative or nonmyeloablative conditioning; pre-existing lung disease; history of lung or chest-wall irradiation; acute post-transplant complications; documented acute or chronic graft-versus-host-disease (GVHD) of the lungs, skin, or other organs; and chest imaging findings on x-ray or computed tomography. For patients with at least two PFTs (n=168), average rate of change of TLC since transplant was calculated. Average monthly change in TLC post-transplant was compared between multiple groups detailed under Results. Analysis of variance or unpaired t-test and linear regression were used for comparison of categorical and continuous variables, respectively. Statistical analyses were completed using BlueSky Statistics 10.3.4, R version 4.1.3. Results: 234 patients were included. Median age at transplant was 51 years (range 18-74), and 163 (70%) were male. Average BMI at transplant was 29.6 (17.9–51.4). 107 (46.3%) underwent myeloablative conditioning, and 124 (53.2%) underwent reduced intensity or nonmyeloablative regimens. The most common indications for transplant were AML (n=87), MDS (n=41), and ALL (n=40). Most had no pre-existing lung disease (n=131, 56.2%), and 61 (26.1%) had evidence of restriction on pre-transplant PFTs. About half (n=119, 51.1%) had a history of lung or chest-wall irradiation. Within 30-days post-transplant, 55 (23.5%) had pulmonary complications (pneumonia or other hypoxic respiratory failure). 151 (64.8%) had either acute or chronic skin GVHD; 91 (39.1%) carried diagnoses of pulmonary GVHD. Most common associated imaging findings were infiltrates (n=120), scarring/atelectasis (n=73), pleural effusion (n=62), and air-trapping (n=50). Mean rate of change in TLC following transplant among all subjects was -0.0473 L/month. Comparative analysis revealed no statistically significant differences in mean values of average change in TLC between those with myeloablative and nonmyeloablative regimens (-0.0733 vs -0.0227 L/month, p = 0.176); pulmonary GVHD or not (-0.0752 vs -0.0203 L/month, p=0.141); GVHD or not (-0.0514 vs -0.0373 L/month, p=0.731); or with or without prior chest irradiation (-0.0234 vs -0.0687 L/month, p=0.226). Women had, on average, positive change in TLC compared to men (+0.00963 vs -0.0709 L/month, p=0.049). Decline in TLC was faster in overweight or obese patients (p=0.006). Conclusion: Our study establishes and describes a cohort of patients with restrictive lung disease post-allo-SCT, a poorly understood and incompletely characterized, yet highly morbid, set of diverse entities into which our findings provide insight. Based on the descriptive analysis we completed, patients experience decline in TLC due to a variety of processes – infiltrative pathology, pleural effusions, and chest wall weakness/restriction. Rate of decline increases with BMI and male gender. Diagnosis of pulmonary GVHD does not seem related to progression of restriction. As more long-term data is obtained from screening PFTs, examining larger datasets may allow for more granular analysis of the numerous possible etiologies.

Abstract Introduction: Telomere biology disorders (TBDs) are caused by germline pathogenic or likely pathogenic variants in telomere maintenance genes and result in multisystem manifestations, including impaired bone health. Patients with TBDs experience reduced bone mineral density (BMD), avascular necrosis, and increased rates of fragility fractures. However, the underlying mechanisms of these bone complications remain poorly understood, and no TBD-specific treatments for bone health currently exist. Our objective was to characterize the clinical features, management, and outcomes of TBD-related bone health issues, including fractures and reduced BMD. Methods: We conducted a retrospective, multicenter, international cohort study using data from 232 patients with TBDs enrolled across 12 centers participating in the Clinical Care Consortium of Telomere-Associated Ailments. Medical records were reviewed for fracture history and BMD assessments by dual-energy X-ray absorptiometry (DXA). The primary endpoint was fracture incidence; secondary endpoints included BMD assessment and fracture management. Data were stratified by pediatric (≤18 years) and adult (>18 years) age groups. Analyses were based on data availability; specific denominators are provided where applicable. Result: Data were available for 53 (23%) pediatric and 179 (77%) adult patients. TERT mutations were the most frequently identified variants in both groups (28% pediatric, 23% adult). Fractures occurred in 25% (13/53) of pediatric patients and 35% (62/178) of adults. In contrast, fractures are estimated to occur in 10-15% of the general pediatric population. The median age at first fracture was 8.2 years (range 2.5–17) in children and 29.2 years (range 18.3–66.9) in adults. Among those with fracture data (n=20 pediatric, n=22 adult), fragility fractures were more common in adults (57%) versus predominantly traumatic fractures in children (61%). Multiple fractures occurred in 65% of affected pediatric patients and 45% of affected adults. Fracture management also differed: non-surgical treatments were used in 60% of pediatric fractures (n=12/20) versus 36% (n=8/22) of adult fractures; surgical interventions were more common in adults (36%). The inheritance pattern (autosomal dominant versus recessive or X-linked) was not associated with fracture risk but the subset of patients with CTC1 mutations were associated with increased fracture risk (n=5 of 124 patients with fracture history, 4% vs n=2 of 247 patients with no fracture history, 0.8%; p=0.04). We also assessed whether vascular manifestations were associated with fracture risk comparing patients with and without fractures. Pulmonary arteriovenous malformations (AVMs) were significantly more common in patients with fractures (17%, 7/42) versus those without (5.4%, 6/112; p=0.04). No significant differences were found in rates of retinal vasculopathy or hemorrhage (13% vs 7.2%, p=0.23) or gastrointestinal bleeding (22% vs 16%, p=0.35). DXA data were available for 171 patients (22 pediatric, 149 adult). The median age at first DXA was 11.4 years (range 4–17.7) for children and 58.2 years (range 20.8–79.4) for adults. Among pediatric patients, the median lumbar spine Z-score was –1.2 (IQR –3.0 to –0.2, n=18), with 39% (7/18) showing a Z-score < –2. Among adults, the median femoral neck T-score was –1.1 (IQR –1.7 to –0.5, n=118), with 21% (29/138) demonstrating T-scores < –2, indicating low BMD. Conclusions: Fractures and reduced BMD are common in both pediatric and adult patients with TBDs, with a substantial burden of early-onset and recurrent fractures. Our findings underscore the importance of early bone health screening and surveillance in TBDs and suggest a possible link between vascular abnormalities and skeletal fragility in this population that requires further study.

Abstract Introduction: Inherited & acquired bone marrow failure syndromes (BMFS) have “stress erythropoiesis” with macrocytic anemia, elevated fetal hemoglobin (HbF) & erythropoietin (EPO) levels. The degree of HbF elevation & its capacity to discriminate between patients with inherited & acquired BMFS needs to be systemically studied. Genetic testing is often delayed & it is important to establish an early clinical diagnosis to make clinical decisions such as initiation of immunosuppressive therapy. Methods: After institutional review board approval, we compiled a clinically & genetically well-annotated cohort of patients with BMFS evaluated at Mayo Clinic. All patients were required to have at least cytopenia(s) & evidence of low or variable bone marrow cellularity upon a bone marrow evaluation. Patients with pre-clonal or malignant disorders were excluded. Clinical diagnoses were independently verified for accuracy. Detailed clinical & laboratory features were recorded. Germline genetic testing was performed as clinically appropriate on peripheral blood or bone marrow samples through targeted panels. Distribution of continuous variables was statistically compared using nonparametric (Mann-Whitney or Kruskal-Wallis) tests, while nominal variables were compared using the Chi-Square test. Kaplan-Meier estimate of overall survival (OS) was computed from the date of diagnosis to date of death or last follow-up (also censored at the time of allogeneic hematopoietic stem cell transplant). Results: Between years 1998 - 2023, 129 patients met criteria for inclusion in the study. Among these, clinical diagnoses included immune-mediated aplastic anemia (AA, n=56, 43%, 21% severe or very severe as per the Camitta criteria), telomere biology disorders (TBD, n=37, 29%), Fanconi anemia (n=10, 8%), GATA2 haploinsufficiency (n=8, 6%), Diamond-Blackfan anemia (DBA, n=2, 2%), deficiency of adenosine deaminase-2 (DADA2, n=2, 2%), congenital dyserythropoietic anemia (CDA, n=1, 1%), congential amegakaryocytic thrombocytopenia (CAMT, n=1, 1%), Schwachman-Diamond syndrome (n=1, 1%) & undefined (n=11, 9%). Among the inherited group, 63% had an identifiable germline pathogenic variant (most commonly, TERT/TERC 30%, FANC group 25% & GATA2 15%) with the rest diagnosed on a clinical basis. Elevated HbF >1% was noted in 79% of all BMFS patients; 83% AA, 70% TBD, 50% in DADA2 & 67% in undefined patients (no data was available for other inherited subtypes). In patients with an elevated HbF, absolute monocyte count (p=0.04) was significantly higher; however, there were no other significant associations with variables at diagnosis such as hemoglobin (p=0.5), white blood cell count (WBC, p=0.14), absolute neutrophil count (p=0.5), platelet count (p=0.7), mean corpuscular volume (MCV, p=0.2), absolute lymphocyte count (p=0.1), absolute reticulocyte count (p=0.4), EPO level (p=0.3), presence of a PNH clone (p=0.2), or presence of a germline variant (p=0.3). When diseases were grouped into inherited & acquired (AA) subgroups, median HbF was not significantly different (median%, 4.5 versus 2.3, p=0.4), with elevated HbF (>1%) in 75% versus 83% (p=0.6) patients respectively. On the other hand, minor PNH clone (>0.1% in at least two cell types) was detectable in the acquired group only (0 versus 55%, p<0.0001). Similarly, acquired group had a shorter (<3 months) onset of cytopenias preceding diagnosis (P<0.0001), with no significant differences in MCV (p=0.2), marrow dysplasia (p=0.09) or atypia (p=0.6), & splenomegaly (p=0.9). At median follow-up of 7 years, Kaplan-Meier estimate of transplant-censored median OS was lower in inherited versus acquired group (16 years versus median not reached, p=0.02). Elevated HbF did not significantly predict adverse OS in either group. Conclusions: Fetal hemoglobin is elevated in both patients with inherited & acquired (aplastic anemia) BMFS. When compared to established variables, elevated HbF has neither diagnostic utility nor prognostic relevance in predicting adverse outcomes. On the other hand, onset of cytopenias and presence of a minor PNH clone are both useful clinical variables in differentiating patients with inherited from acquired BMFS. Our data offers practical clinical utility in making early treatment decisions for patients with BMFS.

Abstract Purpose. Malignant histiocytoses (MHs) including Langerhans cell sarcoma (LCS), interdigitating dendritic cell sarcoma (IDCS), and histiocytic sarcoma (HS) are rare, aggressive neoplasms of the macrophage-dendritic cell lineage. Prognosis is generally poor, particularly in disseminated disease. While mutation-specific therapies are increasingly used in indolent histiocytic neoplasms, their role in MH remains unclear due to the frequent presence of complex mutational profiles. Our understanding of the genomic landscape of MH is limited. This study aimed to comprehensively characterize the molecular alterations in MH.Methods. We conducted a retrospective chart review of MH patients seen at Mayo Clinic and the University of Alabama at Birmingham who underwent comprehensive next-generation sequencing (NGS), including DNA ± RNA analysis. A systematic literature review using PubMed identified additional MH cases with available molecular data.Results. We identified 17 institutional MH cases (3 IDCS, 3 LCS, 11 HS) and 72 cases from the literature (6 IDCS, 11 LCS, 55 HS) with NGS data. Among the 89 total patients, 26 (29%) had DNA-only sequencing and 63 (71%) had combined DNA/RNA sequencing. Twelve patients (n=12; 13%) had a single mutation; the majority had multiple alterations. The most frequently mutated genes were: RAS (n=26; 29%), CDKN1/2 (n=26; 29%), MAP2K1 (n=15; 17%), PTEN (n=11; 12%), BRAF non-V600E (n=8; 9%), PTPN11 (n=8; 9%), CREBBP (n=8; 9%), TET2 (n=8; 9%), DNMT3A (n=7; 8%), fusions (n=7; 8%), BRAFV600E (n=6; 7%), and NF1 (n=4; 4%). Fusions included BRAF::NRF11, BRAF::CLIP2, BRAF::MBP, EGFR::CYTOR, CDKN2A::ADGRG6, IGH::BCL6, and CLIP2-BRAF. Two patients had no alteration detected. Non-V600E BRAF mutations included D594G (n=3), G596R (n=1), G466A (n=1), N581T (n=1), N581I (n=1), and N581S (n=1). Common co-mutations with BRAF included CDKN2A/B (n=9), TP53 (n=6), KRAS/NRAS (n=5), and PDGFR (n=1). Only 2 of 15 MAP2K1-mutated patients (13%) had isolated mutations; the rest had co-mutations, most frequently TP53 (n=3) and BCL2 (n=3). MAP2K1 mutations included F53I (n=2), C121S (n=2), I103N (n=1), Q56P(n=3), E203K (n=1), F53-K57del (n=1), and F53-Q58del (n=1).NRAS mutations (n=6) included Q16K (n=2), G12C (n=1), Q61R (n=1), and 2 unspecified. KRAS mutations (n=20) included G13D (n=3), Q61H (n=3), G12V (n=2), K117N (n=1), V14I (n=1), and 10 unspecified. The most frequent co-mutational pattern (24%) was RAS plus another gene: BRAF (n=3), TP53 (n=6), MAP2K1 (n=2), PTEN (n=4), and CDKN2A/B (n=6).Twelve patients (12/26; 46%) were treated with targeted therapies. Responses included 6 partial and 5 complete responses. Median duration of response was 27.5 months (range 7–54). The sole non-responder had co-occurring NRASand TP53 mutations and received trametinib. Responders received a range of agents matched to their mutations: trametinib (BRAFV600E, MAP2K1F53I, KRASQ16H/PTEND24G/EPHB1R327C, MAP2K1F53I/BCL2/CDKN1B/KIT, and CALR), dabrafenib + trametinib(BRAFG596R/KRASQ61H), imatinib (MAP2K1C121S/METG28881T/DUSP2G137D/HIST1H3BC97Y/GRIN2AS1216C and PTPN11/STK11/GNA11/JAK2), chidamide + sintilimab(IDH2G515A/RHOAG50T/TET23344delC), sirolimus(PTEN/FGFR3/SETD2) and sorafenib (BRAFD594G/KRASK117N/TP53). Most responders (8/11) had multiple mutations; only 3 had single alterations, suggesting multi-mutant MH may still be targetable. Common complex patterns included BRAF+CDKN2A + CDKN2B (n=4), BRAF + CDKN2A + TP53 (n=2), and BRAF + CDKN2A + KRAS (n=2).Conclusion. The MAPK pathway (BRAF, MAP2K1, RAS) is the most altered signaling axis in MH, often accompanied by mutations in tumor suppressors and cell cycle regulators such as TP53, CDKN2A/B, and PTEN. Most MH patients exhibit multiple genomic alterations. Despite this complexity, durable clinical responses to targeted agents are achievable, even in heavily mutated cases. These findings support the use of comprehensive genomic profiling in MH to identify therapeutic targets and guide precision treatment strategies.

Abstract Introduction: The DEAD-box helicase 41 (DDX41) gene, located on chromosome 5q35, is a tumor suppressor gene involved in RNA processing. Germline mutations in this gene can predispose patients to myeloid malignancies. Studies have suggested a more favorable prognosis in those with germline mutations in DDX41, with progression of disease being a frequently encountered cause of death. In this retrospective study, we explored the causes of death in a cohort of patients with pathogenic DDX41 mutations. Methods: This was a retrospective study of patients diagnosed with myeloid neoplasms between 2009-2024 and found to have DDX41 mutation (DDX41m) via next generation sequencing (NGS) at our institution. Myeloid NGS was obtained at or after the diagnosis of myeloid disease. Germline testing (when available) was performed through buccal/skin samples. Carriers of DDX41m without a myeloid neoplasm diagnosis and patients with only DDX41m VUS were excluded. Baseline characteristics at the time of myeloid neoplasm diagnosis included age at diagnosis, DDX41m variant allele frequency (VAF), blood counts (hemoglobin, white blood cell count, platelets), cytogenetics, bone marrow blast percentage, number of co-mutations, and treatment received. Unadjusted comparisons were made using continuous or categorical variables. Median overall survival (OS) was calculated from the time of diagnosis with myeloid neoplasm to the time of death, using BlueSky Statistics V.10.3.4. Results: Of the 112 patients identified to have pathogenic DDX41m, 29 (26%) patients were deceased at the time of data analysis. In this cohort of 29 patients (median age at diagnosis 72 years, 17 (59%) males), disease groups at initial diagnosis included 17 (59%) patients with MDS, 9 (31%) AML, and 3 (10%) MPN. Fourteen of the 17 (82%) of the patients with MDS were classified as MDS-EB1 or MDS-EB2. Seven of the 17 (41%) patients with MDS progressed to AML at the time of death, with 57% of the patients having more than 30% bone marrow blasts at time of progression. Blood counts at diagnosis in the cohort included a median hemoglobin of 9.8 g/dL, platelets of 83 x 109/L, and white blood cell count of 2.1 x 109/L. Twenty-one (72%) of patients had normal cytogenetics. Nine (31%) patients had both a pathogenic mutation and a variant of uncertain significance (VUS) in DDX41, with the remaining 20 patients harboring only a pathogenic DDX41 mutation. Eight (27%) patients underwent germline testing, with 7 confirmed to harbor a germline DDX41m. In the deceased cohort, 9 (31%) patients underwent HSCT. Three of the patients with AML and 1 patient with MDS were observed and did not undergo treatment. Median overall survival of the deceased cohort was 29 months. The median time to AML progression in the group with MDS was 14 months. Causes of death included infection in 11 (38%) patients, 9 (31%) with disease progression/relapse, 3 (10%) with acute gastrointestinal graft versus host disease (GVHD), and 1 (3%) from cerebral bleed. The cause of death was unknown in the remaining 5 (17.2%) patients. Infections included pneumonia, bacteremia, neutropenic enterocolitis, fungal, and unknown etiology. Of the 4 patients who were observed, 2 died from unknown cause and 2 died from disease progression. Of the 16 patients with AML at time of death, 9 (56%) died from disease progression or relapse, 3 (19%) from infection, 2 (13%) from GVHD, and 2 (13%) from unknown causes. Of the 10 deaths in the MDS group, 7 (70%) were from infection, 1 (10%) from GVHD, 1 (10%) from cerebral bleed, and 1 (10%) from an unknown cause. Three cases of infection in the MDS group occurred during remission. Infection was a more common cause of death in the MDS group compared to the AML group, where disease progression prevailed as a cause of death (70% vs 19%, p=0.01). Eight (80%) of the patients with MDS who died from infection were neutropenic prior to death, and 4 (50%) of those patients were on neutropenic antimicrobial prophylaxis. Conclusions: In this cohort of 29 deceased patients with a DDX41-mutated myeloid neoplasm, we characterized causes of death. Disease progression/relapse was a significantly more common cause of death in those with AML, compared to infection in patients with MDS. Our work underscores the need for additional studies looking into non-leukemic causes of death in patients with DDX41m myeloid diseases and potential strategies to mitigate infection-related mortality in this population.

Abstract Background Pediatric patients undergoing hematopoietic cell transplantation (HCT) may develop life-threatening complications and need for ICU admission (ICU). They may display more severe and/or unique manifestations of organ dysfunction syndromes, including acute respiratory failure, sepsis and delirium, compared to the general pediatric In this interim analysis of the POEM study, the utility of a p-MODS screening tool as a predictor of endotheliopathies and/or ICU is reported. Methods: NCT05090345 is a prospective multi-center registry/biorepository trial of children and adolescents/young adults (AYA) undergoing HCT to assess the impact of endotheliopathies on morbidity and mortality. Patients undergoing all HCT types are prospectively monitored for 100 days post-HCT for sinusoidal obstructive syndrome (SOS), thrombotic microangiopathy (TMA), capillary leak syndrome (CLS), diffuse alveolar hemorrhage (DAH) and ICU. The risk of p-MODS was assessed within 7 days of HCT admission and weekly for 4 weeks, to dynamically evaluate organ function across 8 systems: respiratory, cardiovascular, renal, hepatic, hematologic, neurologic, immune, and gastrointestinal. Each system was graded on a 0–5 scale based on clinical and laboratory parameters, with total pMODS calculated using the worst score per organ. Descriptive statistics were used to compare p-MODS score between groups with “events” (defined as endotheliopathy or ICU) versus “no events”. The “no event” comparison group was matched 1:1 by diagnosis, age, gender and HCT type. Results: Among all subjects (n=151), the median age was 11.3 (range: 0.2–26.9) years and the majority (64%) male and Caucasian ( 60%) and 20% Black. Most patients (86.8%) underwent allogeneic HCT. Common indications were hematologic malignancy (45%), solid tumors (19%) and hemoglobinopathies (8.6%).Events (n=50) were observed in 35 of 151 patients, including ICU (n=26), SOS (n=19), TMA (n=3) CLS (n=1) and DAH (n=1). Events occurred at a median of 16 (range: –7 to 85) days post-HCT. Median time to ICU was 15 (range: –7 to 82) days post-HCT. The median pMODS score at the time of ICU was 10 (range: 1–22). All three patients with TMA had preceding ICU admissions and 4 of 19 patients with SOS required ICU management. The median time to onset of endotheliopathy was 17 (range: 6–33) days post-HCT. The median pMODS score at the time of event was 11 (range: 7–18), compared to a median of 8 (range: 5–21) the week prior to onset. Patients who developed events exhibited progressively higher scores compared to matched controls starting at week 1 (p < 0.05). This difference became more pronounced over time, reaching statistical significance at week 2 (p < 0.001). Since physiologic decline may precede overt events, we next analyzed pMODS score change between Day 0 and week 1. A significant difference was observed with median increase of 2 (range: –2 to 16) in the event group versus 1 (range: –3 to 7) in the control group (p = 0.01). Analysis of individual organ system scores revealed notable differences between the event and control groups by Week 1, prior to median event onset. The event group had higher rates of indicators of organ dysfunction in 6 of 8 organ system. The event group had more subjects with acute kidney injury, elevated delirium scores, need for respiratory support, increased bilirubin, moderate to significant weight gain and infections. However, rates of neutropenia and lymphopenia were comparable between groups. Conclusion In this interim analysis, a significant increase in pMODS between Day 0 and week 1 post-HCT may reflect early physiologic decline, with notable changes in respiratory status, acute kidney injury, elevated CAPD scores, and infectious burden. These findings highlight the potential utility of pMODS as an early screening tool to identify pediatric HCT recipients at risk for clinical deterioration. Prospective and dynamic monitoring of pMODS may support timely intervention and improve outcomes in this vulnerable population.

Abstract INTRODUCTION Allogeneic hematopoietic cell transplantation (alloHCT) is indicated in patients (pts) with FLT3-ITD mutated AML. The randomized BMT CTN 1506 trial (Levis et al. JCO) demonstrated that maintenance with FLT3 inhibitor gilteritinib (gilt) significantly improved relapse-free survival (RFS) among pts with detectable peri-alloHCT minimal residual disease (MRD) but was also associated with a greater degree of myelosuppression and infections. In older pts, aged >60 years, increasing post-HCT toxicity may adversely impact outcomes. On this basis, we performed a post-hoc analysis of pts aged > 60 years enrolled onto BMT CTN 1506 to examine the impact of gilteritinib maintenance in older pts. METHODS Pts ≥60 years randomized on BMT CTN 1506 (gilt=47, placebo=57; N=104) were included. The primary endpoint of this post-hoc analysis was RFS. We additionally assessed overall survival (OS), incidence of relapse, non-relapse mortality (NRM) in all pts and in pts with detectable MRD. RESULTS Baseline characteristics were balanced. Age was similar (65.2 vs 64.7 years; p = 0.12), as was prior TKI exposure, and NPM1 mutation status. Myeloablative conditioning (MAC) was used in 8 (17%) pts randomized to gilt and 18 (32%) pts randomized to placebo (p=0.08). Median f/u of survivors was 43.8 months (range:1-61 mo.). Estimated RFS and OS at 48 months were not different between groups (RFS: gilt: 57.0%, 95%CI 44.1-73.7%, placebo: 57.1%, 95%CI 45.4-71.8%; OS: gilt: 56.9%, 95%CI 42.9-46.7%, placebo: 58.4%, 95%CI 46.7-73.1%). While gilt assignment did not affect survival, several predictors of poorer survival were identified including use of reduced intensity conditioning (RIC) (HR = 3.46; p = 0.014), use of calcineurin inhibitor/MTX for graft-vs-host disease (GVHD) prophylaxis (HR = 2.29; p = 0.041), high-risk cytogenetics (HR = 2.64; p = 0.021), and MRD (+) pre-HCT (HR = 2.29; p = 0.018). Finally, relative to gilt assignment, opposing rates of relapse and NRM were observed. The incidence of relapse was lower with gilt (16%, 95%CI 6.8-28% vs 32%, 95%CI 20-45%, p=0.046) but NRM was higher (27%, 95%CI 15-41% vs 11%, 95% CI 4.3-21%, p=0.033). We next analyzed the 49 pts (gilt=22, placebo=27) with detectable peri-HCT MRD. RFS and OS again were not different (RFS: gilt: 44.4%, 95%CI 27.1-72.7%, placebo: 43.8%, 95%CI 28.4-67.5%, OS: gilt: 49.2%,95%CI, 31.3-77.4%, placebo: 55.6%, 95%CI 39.4-77.8%). The incidence of relapse trended lower in gilt pts (25%, 95%CI, 5.3-44.7% vs 45% (95%CI, 25-63%), p=0.2) while NRM trended higher (31%, 95%CI 12-52% vs 11%, 95%CI, 2.7-26%, p=0.12). Seventeen (36%) patients completed planned 2 year maintenance with gilt compared to 29 (51%) completing therapy with placebo. Median time on gilt was 13.2 months. The most common reason for discontinuation was adverse events (n=19, 40%), of which cytopenias (n=8) and infections (n=4) were most frequent. Non-relapse causes of death among gilt treated pts included infections (n=8), intracranial hemorrhage related to thrombocytopenia (n=1), second cancer (n=1), and Guillain-Barre syndrome (n=1). Most infections (n=5) were attributed to GVHD as an underlying cause. CONCLUSION Among pts aged ≥60 years, gilt maintenance did not improve RFS or OS. Gilt use was associated with reduced relapse regardless of peri-HCT MRD status demonstrating disease effect; however, higher NRM limited overall benefit. In older FLT3 mutated pts, employing strategies to mitigate HCT related toxicity, e.g. alternative GVHD prophylaxis strategies, may better accommodate use of maintenance therapy including gilt. Lower doses of gilt (particularly with azole co-use) may be worth exploring to mitigate drug AEs, improve compliance, thus further improving benefit.

Abstract A systematic approach to identifying novel therapeutic strategies is essential for expanding curative options in hematopoietic malignancies such as acute myeloid leukemia (AML). We have recently established a high-resolution genetic platform that enables de novo therapeutic development through the CRISPR-Tiling Instructed Computer-Aided (CRISPR-TICA) workflow (Nat Struct Mol Biol 2024, PMID: 38316881). Using this approach, we further identified a new therapeutic pocket and its lead compound for AML treatment (Sci Adv 2024, PMID: 38394203). Here, we present an integrative pipeline, CRISPR-TICA.ai, which combines CRISPR-based functional genomics with generative artificial intelligence (GenAI)-driven ligand modeling to perform high-throughput evaluation of the top 100 leukemia-essential genes and their therapeutic potential in AML. We first used the CRISPR screen dataset published by Novartis (Cancer Discov 2016, PMID: 27260157), which tile-scanned 68 genes in three cancer cell lines, as the initial training dataset to establish the CRISPR-TICA.ai pipeline. Structurally, ligandable protein surface pockets were identified by employing the P2Rank algorithm (J Cheminform 2018, PMID: 30109435). For each pocket, we used the structure-guided diffusion model DiffSBDD (Nat Comput Sci 2024, PMID: 39653846) to generate previously unseen chemical structures that differ substantially from existing drugs, thereby enabling the exploration of novel chemical space. The model was further fine-tuned using direct preference optimization (NeurIPS 2023) against a curated dataset of known inhibitors to enrich for favorable binding profiles (low binding free energy, ΔG) and calibrated against drug-likeness (QED), synthetic accessibility (SA), lipophilicity (logP), and Lipinski's rule compliance. Additionally, our pipeline integrates ligand–protein contact maps with CRISPR-hypersensitive positions to assess spatial interactions, prioritizing ligands that effectively mask biologically essential residues with high affinity. The utility of CRISPR-TICA.ai in AML drug discovery was evaluated- we selected the top 100 leukemia-essential genes from the DepMap database (https://depmap.org/portal/) and generated custom CRISPR libraries for high-resolution gene tiling screens. These libraries comprised 20,233 sgRNAs to tile-scan all 100 genes (47,901 amino acid residues) in Molm13 AML cells at an average resolution of 2.4 amino acids per sgRNA. This dataset provided in-depth structural and functional input for CRISPR-TICA.ai to examine AI-enabled ligand design for over 300 high-confidence druggable pockets in AML. We further developed a user-friendly, web-based 3D visualization platform that interactively maps functional CRISPR hotspots alongside predicted pockets and ligands to enhance accessibility and reproducibility. This tool allows visualization of CRISPR tiling scores, predicted binding pockets, and ligand prioritization, enabling interactive exploration and application. To annotate therapeutically relevant sites, we incorporated data from multiple genomic and proteomic resources, including AlphaFold protein models, Pfam domain annotations, post-translational modification (PTM) sites, and COSMIC leukemia mutational data. Exportable files and GitHub integration are also provided for advanced users to deploy CRISPR-TICA.ai within their facilities.In summary, our study introduces a scalable, leukemia-focused framework for de novo therapeutic discovery. By integrating high-density CRISPR mutagenesis with diffusion-based GenAI modeling, CRISPR-TICA.ai provides a rational roadmap for compound design targeting residues essential for leukemia maintenance. Importantly, the CRISPR-TICA.ai pipeline is adaptable to CRISPR tiling datasets from solid tumors (e.g., the Novartis dataset), highlighting its versatility across diverse cancer types. This approach is especially promising for addressing historically undruggable and resistance-prone targets in AML and beyond.

Abstract Introduction: Iron deficiency (ID) is among the five leading causes of years lived with disability in women of reproductive age worldwide and affects >50% of pregnancies. Both ID and its terminal sequela, iron deficiency anemia, are associated with adverse maternal-fetal outcomes including postpartum hemorrhage, impaired fetal neurocognitive development, preterm labor, and maternal mortality. At present, no United States (US) guidelines exist for the universal screening of ID in pregnancy. Moreover, diagnostic thresholds for ID established by the Centers for Disease Control and Prevention (CDC; <15µg/L ferritin in all trimesters) and the World Health Organization (WHO; <15µg/L ferritin in first trimester and no guidance in second or third) fall below more sensitive physiologically-based thresholds of no less than 20-25µg/L ferritin in pregnancy. As such, ID screening and treatment represents an enormous gap in the health of expectant mothers and in which intravenous iron supplementation has been established as an effective and safe treatment option. We sought to address this gap by conducting the first cost-effectiveness analysis of screening ferritin thresholds for ID diagnosis and treatment in pregnancy. Methods: We built a Markov simulation to examine the cost-effectiveness of screening all individuals for iron deficiency in the second and third trimesters of pregnancy with the following strategies: 1) ferritin threshold <30µg/L, 2) ferritin threshold <15µg/L, and 3) status quo (i.e., no screening). The analysis was conducted over a lifetime horizon, from the US health system perspective and across all accepted willingness-to-pay (WTP) thresholds measured in $ per quality-adjusted-life-year (QALY). Costs were assessed in 2025 USD, inclusive of age- and sex- specific annual healthcare costs informed by the US Medical Expenditures Panel Survey. Age-specific fertility rates were sourced from the US National Vital Statistics Reports and used for model internal validation. The trimester-specific prevalence of ID was informed by the largest published longitudinal prospective cohort study of iron status in pregnant women to date. The probability of iron-related adverse events including anaphylaxis was sourced from the World Health Organization Vigibase data. Aligning with prospective data of ID in pregnancy, women entered the model at age 15 and were screened (or not screened) for ID per treatment strategy during the second and third trimester of each pregnancy in their lifetime. If found to be iron deficient, patients were treated with a single dose of 1g intravenous iron dextran. A separate scenario analysis from a societal perspective also incorporated the cost of wages lost to infusion time. Effectiveness was measured in QALYs and parameterized multiplicatively using age-, sex-, pregnancy-, and ID-specific utility values, exclusively assessing maternal benefit in pregnancy without accounting for any fetal benefit. The primary outcome of interest was the incremental cost-effectiveness ratio (ICER). We conducted deterministic and probabilistic sensitivity analyses, capturing uncertainty across all parameters simultaneously over 10,000 Monte Carlo iterations. Results: Internal model validation demonstrated an average fertility rate of 1.62 [95% credible interval (CI) 1.59-1.65], in line with the US fertility rate of 1.62. Screening for ID with a ferritin threshold of 30µg/L versus 15µg/L versus no screening accrued $213,000, $212,800, and $212,400 and 25.12, 25.11, and 25.09 QALYs per individual, respectively. Screening at a ferritin threshold of 30µg/L was the cost-effective strategy in 100% of 10,000 Monte Carlo iterations, with an ICER of $23,000/QALY [95% CI $20,000-$27,000/QALY], below all accepted WTP thresholds. Screening at 30µg/L remained the cost-effective strategy from a societal perspective, at an ICER of $33,000/QALY [95% CI $29,000-$38,000/QALY]. Deterministic sensitivity analysis revealed that no parameter variations changed this conclusion.Conclusion: The identification and treatment of ID beginning in the second trimester at a ferritin threshold of <30µg/L is the cost-effective intervention, in comparison to a threshold of <15µg/L or no screening. These results fill a critical gap in maternofetal medicine and may help inform the forthcoming American Society of Hematology Clinical Practice Guidelines on the Diagnosis and Treatment of Iron Deficiency.